By: Katherine Dick and Andrew Briggs, DPhil
Avalon Health Economics
April 16, 2020
An important and growing field within health economics and outcomes research (HEOR) is what we call “HEOR Analytics.” Traditional HEOR has focused mainly on model building and utility development, with the end goal of calculating costs per quality-adjusted life years (QALYs). But as data sources have become more nuanced and complex, and product value pathways have become more intricate, there is a need to push the limits of HEOR by applying state-of-the art statistical analyses of data. These tasks include novel statistical analyses of HEOR data, such as trial data or quality of life data, and novel applications of those analyses to statistical and economic models. In this blog entry we report on a good example of HEOR analytics: our recent study of covariate-adjusted analysis of trial data of lenvatinib versus sorafenib in hepatocellular carcinoma.
Hepatocellular carcinoma (HCC) is the most common type of liver cancer and a leading cause of cancer-related deaths worldwide.[1, 2] Chronic liver disease and cirrhosis are critical risk factors in the development of HCC. Sorafenib is currently the only first-line systemic treatment available to treat unresectable HCC, but lenvatinib showed potential as an alternative therapy in a Phase 2 clinical trial. The Phase 3 REFLECT trial found that lenvatinib was noninferior to sorafenib in the primary outcome of overall survival and superior in the secondary outcome of progression free survival.
The Phase 3 trial was randomized, but there was a notable imbalance in baseline characteristics between the trial treatment arms. These baseline variables were important prognostic factors for overall survival, so the imbalance appeared to bias the outcomes against lenvatinib. Our covariate-adjusted analysis of the trial data assesses the magnitude of and adjusts for these imbalances.
To determine the potential importance of the candidate variables identified by the clinical experts, each variable was entered into the Cox proportional hazards regression model as a univariate adjustment of the treatment effect. The Forest plot in Figure 1 shows the univariable impact on the estimated hazard ratio for lenvatinib treatment compared to sorafenib after adjusting for each covariate. In terms of these univariable results, MPVI or EHS or both, AFP < 200 ng/mL, disease site, hepatitis B etiology, and receipt of a previous procedure are all predictive of overall survival and adjusting for them influences the estimated hazard ratio of the treatment effect in favor of lenvatinib.
A multivariable adjusted analysis was developed using a “forward stepwise” procedure to systematically select covariates for inclusion. The chosen multivariable Cox model analysis resulted in an estimated adjusted hazard ratio for lenvatinib of 0.814 (95% CI: 0.699–0.948) when only baseline variables were included. Adjusting for post-randomization treatment variables further increased the estimated superiority of lenvatinib. This analysis suggests that the original noninferiority trial likely underestimated the true effect of lenvatinib on overall survival due to an imbalance in baseline prognostic covariates and the greater use of post-treatment therapies in the sorafenib arm. To read more, the full text of the article is available from the British Journal of Cancer at https://www.nature.com/articles/s41416-020-0817-7.